View
0
Download
0
Category
Preview:
Citation preview
AD-AI38 843 REAL TIME IMPLEMENTATION OF NONLINEAR OPTICAL i/iPROCESSING FUNCTIONS(U) HUGHES RESEARCH LABS MALIBU CA
BHSOFFER DEC 83 AFOSRTR- 84-0141 F49620-81-C-0086UNCLASSIFIED F/G 20/6 NL7miEiEEBEiII
IBhBhBhhhBhhhE
~~~~~3 2.5~ I 4. 111I.- IuI~****~ 11111.
1.54.
MICROCOPY RESOLUTION TEST CHART
NATONAL BUPEAU OF STANDARDS 191.1-A
4*4
AFOSR .",. 84- 0 14 1
REAL-TIME IMPLEMENTATION OF NONLINEAR
O' OPTICAL PROCESSING FUNCTIONS
0-B.H. Soffer
Hughes Rese;,rch Laboratories
3011 Malibu Canyon Road
Malibu, CA 90265
December 1983 )
F49620-8 I-C-0086
Annual Technical Report
15 June 1982 through 15 June 1983
This manuscript is submitted for publication with the understandingthat the Uvited States Government is authorized to reproduce and dis.
.5*rpare Fotribute reprints for governmental purposes. E T IC
Prepared For
AIR FORCE OFFICE OF SCIENTIFIC RESEARCH
Boiling Air Force Base
Washington, DC 20332
7re 3rj -of'cti.,.)rs * "L.. u
ureterj as Itcessa' 'V u'e e = J .'. . wA r O ffce -)f , e " "c se3cr
.-. l[ FILE COPY ." . . . - -'... .... ',]f te .
%." ' •::; 0%;: " '::": ' '" " - - /- -" "-'-:.",-_7 ' ", "- . , :, .": '
V°
~UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE (When Data entered)R
' REPORT DOCUMENTATION PAGE BEFORE COMPLETING FORMRGOVT ACCESSION NO. 3 RECIPIENTS CATALOG NUMBERAFOSRTR. 8 4 - 0 141 ' /
4. TISLE (and Subtitle) 5 TYPE OF REPORT & PERIOD COVERED
Real-Time Implementation of Nonlinear Optical Annual Technical Report' Processing Functions 15 June 1982 - 15 June 1983
6. PERFORMING O1G. REPORT NUMBER
7. AU THOR(a) S. CONTRACT OR GRANT NUMSER(s)
B.H. Soffer F49620-8i-C-0086
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROjECT rASK"-" AREA & WORK UNIT NUMBERS
Hughes Research Laboratories AU M
3011 Malibu Canyon Road 3/B/Malibu, CA 90265 ____ _./__
- II. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE
December 198313. NUMBER OF PAGES 1 7 , 3
14 MONITORING AGENCY NAME & ADDRESS(II different from Controling9 Qfice) I5. SECURITY CLASS. fa this report)
Air Force Office of Scientific Research / Unclassified
Bolling Air Force BaseWashington, DC 20332 15s. OECLASSIF1CATION DOWNGRADING
SCHEDULE
4 16. DISTRIBUTION STATEMENT (of this Report)
This manuscript is submitted for publication with the understanding that the
, United States Government is authorized to reproduce and distribute reprints
* . for governmental purposes. A,. . . .,i).lio release;"• "' o •: :,unllimited.
17. DISTRIBUTION STATEMENT fof the abstract entered in Block 20, if different from Report)
• 18. SUPPLEMENTARY NOTES
Research sponsored by the Air Force Office of Scientific Research (AFSC), under
Contract F49620-81-C-0086. The United States Government is authorized to
reproduce and distribute reprints for governmental purposes notwithstanding
any copyright notation hereon.19. KEY WORDS (Continue on reverse side it nee.esary and Identify by block number)
Optical signal processing, optical data processing, signal processing,
data processing, liquid crystal devices.
I
20. ABSTRACT (Continue on reverse stde If necessary end Identify bv block number)
Optical data processing has not yet achieved its potential of increasedcapacity and speed compared with conventional electronic techniques, primaril:
S. for lack of a pratical real-time image modulator, and because optical techniqueshave been almost exclusively limited to linear operations. The continuing
research outlined in this report attacks these issues by studying the implementa-tion of real-time nonlinear parallel-processing techniques. The variousimplementations studied in this program all employed real-time liquid-crvstal
light valves developed and specially modified for these tasks by Hughes Research
SDD I A"73 1473 EDITION OF 1 NOV 6 IS OBSOLETEUNCLASSIFIED
SE - RITY CLASSIFICATION OF THIS PAGE When Dete Daa .red)
.- -UNCLASSIFIEDS ECUIRITY CLASSIFICATION OF THIS PAGE(Wh.* Date Entered;
Laboratories. One approach we investigated early in the program was to modifyand characterize the twisted-nematic liquid-crystal (LC) devices, and then usethem in a coherent optical data-processing apparatus using special half-tcnescreen masks, custom designed for special functions at USC in a cooperativeeffort under an AFOSR grant. Using the half-tone mask technique, we demon-strated logarithmic nonlinear transformation, permitting us to simplifymultiplicative images and perform homomorphic filtering. Furthermore, a novelanalog-to-digital converter based on a modified pure birefringence LCLV wasdeveloped. It can perform real-time parallel processing using incoherent light,
.,.-.. [and it promises high data throughput rates. In addition, a novel device thatconverts light intensity variations to LC grating period variations was fabri-cated, and is currently being evaluated and improved. This device permitsnonlinear functions to be implemented directly without the need for speciallymade half-tone masks. Besides nonlinear analog functions, this variable gratingmode (VGM) device has demonstrated the capability of performing digital logic.Logical functions are merely special cases of nonlinearities. In this periodwe report a novel technique for spoiling the long-range order in the VGM sub-strate to show a remarkable improvement in the temporal response. An alterna-tive to the VGM device, which retains the essential optical processing features,is described.
-.I
*J' IA
-o_
UNCLASSIFIED. . . . . . . . . . . . . . . . . . . . . . .. . . .
LIST OF ILLUSTRATIONS
FIGURE PAGE
1. Design of holographic grating substrate .......... 8
2. VGM domains aligned perpendicular tothe horizontal ..................... .............. 9
3. Apparatus for studies of the dynamics ofVGM response .......................... 10
4. VGM temporal response; +20 to +30 V dc,1 um grating substrates, 500 msec/cm ............. 12
5. VGM temporal response to +30 + 40 V dc,I um grating substrates, 2 sec/cm ................ 12
b 6. VGM temporal response; +30 to -40 V dc,1 um grating substrates, 300 msec/div ............ 14
7. VGM domains in cross-rubbed PVA cell, 180X ....... 16
8. VGM temporal response; -40 to +60 V dc, .......... 16
S9. SEM photograph of 0.4 um holographicgrating of ITO on glass ........................ 17
10. VGM temporal response; +40 to +30 V dc,0.4 urn grating substrates perpendicularlydisposed, 2 sec/div .............................. 17
11. VGM frequency response; -40 to +20 V, 1 urmgrating substrates, 1 Hz to 30 Hz sweep .......... 18
12. V- %ped grooves with (111) walls areproduced by etching in (100) silicon slice ....... 23
13. Blaze angle is determined by inclination:9 "- angle between the silicon surface and the
(111) plane ...................................... 23
"" 14. Short period grating in single crystalsilicon .......................................... 24
AIR FORCE 0FFICE DF SCIENTIFIC RESEARCH (APSC)
A T EW J. .....Chief, Technical Information Division
VOT.,..
..................................................................................................
II
- '-. -.- ---...
iSECTION 1
INTRODUCTION
- For the past two decades optical data processing (ODP) has
IL promised a vast increase in processing capacity and speed over
conventional electronic techniques. This promise has never been
fulfilled for several reasons, most notably because of the lack
of a practical real-time image modulator, or light valve, and
because optical techniques were almost exclusively limited to
linear operations. These restrictions have been removed by the
development of the liquid-crystal light valve (LCVL) by Hughes
Research Laboratories (HRL), and by nonlinear parallel-
processing techniques developed by the University of Southern
California (USC). Thus, it is important to determine how
successfully non-linear parallel-processing techniques can be
implemented in real time with the various LCLVs.
The implementation and evaluation of these techniques have
a direct relationship to current Air Force technology. Perti-
nent Air Force interests include multidimensional real-time
signal and image processing with varied applications, including
nonlinear filtering for trajectory control and guidance, "smart"
rsensing, picture processing, and bandwidth compression. These
technologies could benefit substantially from the increased
processing capacity and speed that the proposed research nay
yield.
-.Here we describe nonlinear optical transformation methods
and the motivation for studying a real-time application of those
techniques. In Section 2 we describe the progress made during
the current program year. The HRL-developed photactivated
twisted-nematic LCLV, the variable grating mode (VGM) LCLV modi-
fication and its system configuration for optical processing,
and optical logical operation has been described in previous
reports. A study of the dynamic performance of the VGM LCLV has
-. A
11
:, ;,~~~~~~~~~~~~~~~~~~.. -'...-... -..... -.... \ ..........-..-......... ,. . ...... •.......... - ... --.-......... '........... .
been performed in this period using a novel technique to reduce
the time of VGM domain formation. Empirical modeling of the VGM
effect has also been performed.
The implementation and evaluation of these techniques has a
direct relationship to Air Force interests in the multidirec-
P Itional processing of real-time signals and images, with applica-
tions including nonlinear filtering for guidance and trajectory
control, smart sensing, picture processing, and bandwidth
compression. These technologies could benefit substantially
from the increased processing capacity and speed that these
techniques may yield.
Until now, specified nonlinear operations have been
* performed only with great difficulty. Coherent optical tech-
niques are essentially restricted to linear opeations. Digital
.- ""processing to produce nonlinear transformations is possible, but
only in a slow, serial fashion. Certain nonlinearities can be
" . produced by special photographic techniques, but the speed,
i accuracy, reproducibility, and dynamic range of these techniques
are limited.
*' We have been pursuing different tasks to attempt to over-
come these shortcomings. The first made use of special half-
tone screens to modulate the input image in conjunction with
K - coherent optical proessing to desample in the Fourier plane.
This technique has made it possible to implement nonlineareffects when higher orders of the half-tone diffraction pattern
are examined by spatial filtering. Sawchuk and Dashiell of USC
have shown, using specially fabricated half-tone screens, how a
very wide class of two-dimensional point nonlinear functions can
be implemented with a large dynamic range as a function of
screen design and diffraction orler. The nonlinearities can be
continuous or discontinuous. Operations such as taking loga-
rithms, exponentiation, level slicing, intensity bandstopping,
and histogram equalization can be performed. We have expanded
i L' the half-tone screen technique by substituting a real-time
2
- - -. : -'- - * ~ .*.:* -i -. -.- -- p *.-7- *-7,.-.*. ... * .. . ,- -
photo-modulated LCLV for the static photographic recording
N medium, and we have successfully demonstrated a logarithmic
nonlinear transformation using this technique. This transforma-
tion is useful for homomorphic-filtering applications as we have
demonstrated. In cooperation with USC, we have also studied the
*performance potentials and limitations of this implementation
and how to iteratively modify and improve the LCLV and half-tone
masks.
A second general method which we are presently studying
also overcomes the limitations of serial or photographic proces-
sing; it employs, in one realization, a liquid-crystal effect -
VGM - that can, when incorporated into a new type of photocon-
* ductive structure, automatically map image intensity variations
into positions in Fourier space. Filtering and reconstructing
can then yield many desired nonlinear transformations of the
image without the need for specially constructed half-tone
masks. 4 new additional parameter, the intensity, has thus been
n made available for optical image and data processing. Recogniz-
ing that logical operations are merely a special case of
nonlinear operations, we have demonstrated a unique and highly
advantageous optical computing scheme using the VGM technique.
The VGM device is still in an early stage of development and
much material research and device development is still needed to
make it into a practical, real-time, reliable optical image
modulator. This work is still continuing. Alternative practi-
cal realizations of intensity to positional mapping are also
being studied that may overcome some of the shortcomings of VGM,
yet retain the great flexibility offered by intensity-to-
*. positional coding.
3p-
!
. . .* 7
SECTION 2
PROGRESS DURING CURRENT YEAR
During this program year, we pursued several tasks which
aimed at further improving our understanding of the VGM effect
* and at developing a more practical VGM photoactivated light
valve device. In the VGM effect, certain liquid-crystal ele-
ments exhibit a phase grating whose local period is proportional
to the locally applied voltage. When incorporated into a light
valve structure along with a photoconductive layer, local varia-
tions in the spatial distribution of light signal or image
intensity are converted into local variations in grating period.
In the Fourier transform plane, a new parameter - the intensity-
is coded into the spatial distribution in a way that allows
great freedom to perform nonlinear optical data and image pro-
cessing, and to perform optical computations as well. In this
period, we continued to examine many aspects of the device,
including the temporal dynamic properties of the light valve
using novel schemes to increase the speed of response. The
detailed optical polarization properties of the VGM diffraction
patterns were also studied with the goal of completing the
modeling of the molecular confiquration of VGM. We also studied
alternative schemes to the VGM wherein the main feature of
* - converting intensity variations to positional variations would
* - be included along with the desired faster time response
_ characteristics.
A. IMPROVED VGM RESPONSE TIME
At the end of the last period we began a study which showed
promise of improving the VGM response time by spoiling the long
range order of the periodic domains. In this sub-section we
will review these results and discuss our progress during this
year.
5
From the device standpoint, the '1GM response times, as we
5have noted before, are too slow for many applications. It would
be desirable to have responses compatible with TV frame rates or
faster, on the order of tens of milliseconds. In this section
we describe a novel approach to, and some preliminary results
* for, achieving faster dynamic response. It is based on the
- intuitive notion that the '1GM domains evolve slowly in time,
both along the domain direction and in the perpendicular
direction.
There is some visual evidence for the slow evolution of the
* domains. When the voltage is changed in a '1GM cell, say to a
slightly higher value, under the polarizing microscope one can
* . see the new additional lines coming into the field of view. New
lines form by "unzipping" one or more of the numerous fork-
shaped dislocations which occur throughout the area of the '1GM
structure. The unzipping process typically traverses 20m at a
velocity on the order of magnitude of 10 in/sec.
5 It seemed possible that if the length of the VGM grating
domain could be shortened radically so that the unzipping would
- not have to travel far, that the time of formation of the new
* . VGM steady state could be shortened. One way to cover a two-
p dimensional area with short '1GM domains would be to have the
- longer domains interrupted by a fixed set of parallel interrup-
tions perpendicular to the length of the domains. The interrup-
* tion would have to present a strong electrical and/or mechanical
perturbation.
We constructed a special cell with one narrow 1-mm groove
parallel and another perpendicular to the '1GM domains. No
* difference in response times was noted between these grooves,
* nor between the grooves and the normal large area '1GM cell
geometry.
. Next we fabricated a set of holographically produced grat-
ings with a one thousand times smaller, 1 mn period. These
* gratings were formed on ITO coated glass substrates etched
through the conductive coatings. The conductive strips were all
6
electrically connected at one edge. The design of the substrate
I nis shown in Figure 1. Two orthogonal gratings were put on each
substrate so that both parallel and perpendicular opposing elec-
trode gratings could be configured in the complete liquid crys-
tal sandwich cell sturcture. This assured that both these
* arrangements, and the grating-free areas as well, could all be
studied under the same conditions of thickness, temperature
* . impurities, and aging.
The VGM domains in a working cell 6 m thick with 20 V-dc
applied are shown in Figure 2. The domains are perpendicular to
the two 1- m holographic gratings. In the photograph, that
grating cannot be seen, and only large occasional defects along
the direction of the grooves show up. What is noteworthy in
this typical photograph is the large number of forked disloca-
tions, as compared with a cell made of normal substrates without
any gratings. The domains, however, are not merely 1 m long
(the period of the grating), but typically extend to 10 m in
length. In a typical cell without gratings the domains may
extend several millimeters before forking.
The LC mixture, Merck NPV, was aligned perpendicular to the
direction of the grooves of the holographic grating in those
3 experiments that we will report on here. In another set of
experiments with the alignment parallel to the grooves, no
significant effects were noted.
The temporal response characteristics proved to be complex
and impossible to describe merely with a single set of charac-
teristic rise and decay times. The responses are non-exponen-
tial and multiprocess with a pronounced memory or hysteresis.
There is also a secular degradation in response. All of these
complexities remain to be analyzed.
The apparatus shown in Figure 3 was employed to perform the
" . response time measurements. The measurements were taken on the
first order diffracted light from the VGM cell. The diffracted
spot of the laser beam was from 3 to 5 mm in diameter at the
detector 20 cm from the LC plane. The detector was positioned
7
b'. . . . . .
12712-1
G GLASS
ITOCOATING
Figure 1. Design of holographic grating substrate.
8
r
*12545-1
.1
Figure 2. VGM domains aligned perpendicularto the horizontal; 1 um holographicgrating in a 6-urn thick cell with200 V DC appli.i
9
1 1300-9R2
SOLENOIENROROVALV
ITEMPEATURE APRTUEL
THERMO DETE CTORTO j
13EA
EXANE
in each experiment at the angle corresponding to the steady
U state diffraction angle for the particular field applied. A
mask was positioned in front of the detector to give an angular
resolution of 1 mm at 20 cm from the VGM cell plane.
To measure the temporal response of this cell, the voltage
* on the cell was repetitively switched from one dc value to
another in a rectangular wave pattern. One detector was placed
at each of the diffraction positions appropriate for each vol-
tage. The outputs of these detectors were displayed on dual
traces of an oscilloscope.
For example, in Figure 4 the top trace records the output
of the detector placed at the 30-V diffraction position, while
the bottom trace simultaneously records the output of the detec-
tor at the 20-V diffraction position as the voltage is repeti-
tively switched from 20 to 30 V in a 2-sec period. The motion
of the diffracted light beam from one detector to the other is
complex. Not only does the beam travel from one detector to the
fi other, but it also undergoes a reformation and condensation as
it approaches steady state, making the temporal response very
* complicated, as can be seen from the oscilloscope traces. There
is also a sudden flash of light, largely centered around the
U current position, whenever the voltage is switched. This flashshows up as a strong spike on the traces. From Figure 4 we
estimate the rise time to be 600 200 msec and the delay timeis now only in the tens of milliseconds! In Figure 5 the cell
was refilled and traces were recorded from 30 to 40 V. The top
trace corresponds to 40 V. Here the rise times were on the
order of 100 msec, but the decay times were 600 and 1,700 msec,
respectively, for 30 and 40 V. This was clearly not a reproduc-
tion of the remarkably rapid decay observed in Figure 4. At
this stage in our experiments, we have been able to reproduce
the fast performance only from time to time and in only certain
of the cells. However, we are greatly encouraged to have seen
* such significant improvement in temporal response using this new
technique. For example, after the gradual deterioration in VGM,
12546-2
* Figure 4. VGM temporal response; +20 to + 30 V dc,1 Pmn grating substrates, 500 msec/cm.
12545-3
Figure 5. VGM temporal response; +30 to +40 V dc,
1 vmgrating substrates, 2 sec/cm.
12
VW 7- '-6q
performance of the particular cell studied in Figures 4 and 5
(noted by higher thresholds and increasing conductivity) could
no longer be tolerated and the cell was refilled. The decay
times were four times slower than the original values. Upon a
* later second refilling, the times were even slower. This
phenomenon is not yet understood.
Next, quasi-ac experiments were tried. If operational, ac
would avoid electrochemical poisoning of the liquid crystal.
* . Switching from the same positive to negative voltage values,
e.g., +20 V to -20 V in square wave fashion also produced fast
decay times of the order of 50 msec and rise times of the order
of a second. But as the diffraction angle is the same for the
positive as for the negative voltage, and since there is somememory of previous states at the switching times we emplopyed,
we abandoned that choice of equal voltage as being misleading.
To take advantage of quasi-ac and yet have more meaningful
temporal response measurements, we repetitively performed
switchings from +30 to -40 V. Figure 6 illustrates the VGM
response of a 1 inm holographic grating cell. It should be noted
that unless otherwise stated the configuration normally employed
is with both electrode and counter electrode gratings parallel
to each other. In the top trace the rise time from +30 to -40 V
* is 300 insec and the decay time :001 insec. In the lower trace,measured on the second detector, the rise time from -40 to +30
*,is 1 sec and the decay time from +30 to -40 is z50 insec. The
asymmetry in rise times is not understood.
* . The production of holographic gratings is time consuming
* .and subject to many difficulties and failure modes in the expo-
sure and etching phases. It seemed that it might be possible to
spoil the long range order in the VGM domains by a simpler
method: by producing linear, randomly spaced scratches on the
substrate surface.
One method of liquid crystal alignment, often used for immedi-
ate results, is rubbing. The substrate is rubbed with coarse mate-
rial, perhaps diamond paste, to effect a quasi-linear pattern of
13
12778-1
Figure 6. VGM temporal response; +30 to -40V dc,
1 p.m grating substrates, 300 msec/div.
14
surface forces which align the liquid crystal. There is no
U definite period in these linear scratches. They are randomly
spaced. We first rubbed the surface in one direction and then
* cross-rubbed it in the orthogonal direction. The liquid crystal
successfully aligned in the last direction that was rubbed, but
* the underlying cross-rubbing seemed to achieve the desired
effect of breaking up the long range VGM domain order. This can
be seen in Figure 7, a polarizing microscope photograph of a
cross-rubbed PVA substrate cell spaced at 6 pin and filled with
Merck NPV with 50 V dc applied. The long range order is clearly
* . spoiled. The temporal behavior of this cell is shown in Figure 8.
The top trace -40 to +60 V shows a rise time of 1.9 sec and
*.decay time of -10 msec. Even this crude technique has signifi-
cantly reduced the response time of the VGM device!
A set of 0.4 wm holographic gratings was fabricated and
tested. This period is significantly smaller than the typical
VGM instability period and it was unclear whether, for best
results in improving the response time, the grating period
should be larger or smaller than the VGM period. A scanning
electron microscope photograph of the 0.4-pm grating of ITO
onglass at 40,000 diameters is shown in Figure 9. The results
* with this grating have not been as dramatic as with the 1 pm
grating, but a significant improvement over no grating at all
was still observed. In Figure 10 we display the results for
* perpendicularly disposed grating electrodes and for +40 to
+30 V dc. The rise and decay times are -400 msec. This was
typical of the usual grating arrangement as well.
A 6-pm set of holographic grating substrates were fabrica-
ted into VGM cells and evaluated. This 6-pm period is somewhat
below the typical period of the normal VGM phenomena. With
these substrates we could not produce clear VGM phenomena. The
diffraction was extremely smeared and confused.
To illustrate the frequency response of the VGM phenomena
in the 1 Pm grating cells, we show in Figure 11 (a), (b), and
15
12778-?
Figure 7. VGM domains in cross-rubbed PVA cell, 180X.
12278-3
Figure 8. VGM temporal response; -40 to +60 V dc,cross-rubbed PVA substrates. 2 sec/div.
.16
12778-4
.
-. .Figure 9. SEM photograph of 0.4 um holographicgrating of ITO on glass.
12778-5
S.1.
L Figure 10. VGM temporal response; +40 to +30 V dc,0.4 um grating substrates perpendicu-larly disposed, 2 sec/div.
17
i:: - .-.. .--. .... .... ... ... ..-- . . .. . . . .
1,
12778-6
S (a)
(b)
(C)
IA Figure 11. VGM frequency response; -40 to 4-20 V,1 urn grating substrates, 1 Hz to 30 Hzsweep. (a) Square wave; (b) Sine wave;(c) Sawtooth wave.
18
4- -. 7 . 4.
(c), the -40 to +30 V temporal response using square, sine and
I sawtooth wave excitation, respectively, swept from 1 tc 30 Hz.
There is clearly significant response to beyond 10 Hz. Most
noteworthy is the strong spiking, mentioned earlier, which can
now be seen to be most prominent at a certain intermediate range
of driving frequencies, depending upon whether square, sine, or
sawtooth waves are employed. It would seem possible to drive
devices with these waveforms and implement the rapid response
implied by these experiments.
*'- . B. PHYSICAL ORIGIN OF THE VARIABLE GRATING MODE EFFECT
Obviously, the important VGM device operational properties
depend on the detailed nature of the grating formation anddynamic orientation process. We have, in conjLnction with our
• . colleagues at USC, made some experimental and theortical studies
* . this year to elucidate the fundamental nature of the grating and
of the physical mechanisms that give rise to the observed
I periodic instability and its dynamics.
The grating produced by the periodic spatial re-orientation
* of the nematic liquid crystal molecules is quite unusual and
gives rise to striking polarization-dependent properties in
I optical diffraction. By studying the intensity and polariza-
tion of the diffracted light for each diffracted order, we have
been able to determine the spatial distribution of the molecular
. orientation within the VGM liquid crystal layer. The polariza-
tion dependence of the diffraction is due to the birefringent
. .phase grating in which the orientation, the principal axis of
the index ellipsoid, varies periodically both in the plane of
the grating and normal to the plane. The spatial distribution
-of the ends of the liquid crystal molecules is approximately'I*cycloidal. This was determined by matching the experimental
"" . results with theoretically calculated values for different
possible models. The technical details are described in a
19
preliminary publication reproduced in the appendix. A more
p detailed publication is being prepared as well.
A theoretical study of the spatial distribution of the
liquid crystal molecules is also being carried out by direct
-minimization of the free energy. The preliminary results of
* this study show that the angular dependence observed experimeni-
tally can be predicted theoretically. The origin of the VGM
* effect is still in doubt, but the evidence strongly points to
the converse flexoelectric effect. The details of this theoret-
ical work to date are also reproduced in the appendix.
The dynamics of grating formation, reorientation and relax-
ation are subjects of on-going experimental and theoretical
Q investigation. Understanding of the basic physical principleskQ underlying these effects is vital to the success of efforts to
improve the VGM response time.
C. ALTERNATIVES TO THE VGM DEVICE
U The variable grating mode liquid crystal effect provides
only one possible means of achieving parallel intensity to posi-
tion encoding. The processing potential of the intensity-to-
position algorithmn is so great that we have been exploring alter-
native implementations. one promising possibility is electro-optic beam deflection, as discussed previously. The angle that
light is deflected by refraction from a prism of a small angle is
approximately proportional to the prism angle and the index of
refraction. The index refraction can be varied y electro-
optically in many materials; the effect is particularly large in
LCs where the change in index between the extremes of ordinary and
extra-traordinary polarizations can be as much as 0.2 with rela-
tively small voltages (-10 V). By varying the voltage locally to
conform to a 2-D image which illuminates a photoconductive layer on
top of a prismatic wedge of LC, one would have a device that
transforms local image intensity into local angles of deflection.
20
In the back focal plane of a lens, these angles are mapped into
S displacements, i.e., a spatial Fourier transform space, with the
new additional parameter of intensity included as well, just as
- "in the VGM device. In the last period we demonstrated electro-
optic deflection with liquid crystal material. The problem
E encountered was the very slow response time of thick liquid
, cells. In order to bring the response time to the millisecond
region we conceived of an array of replicated little deflecting
-prisms. Since only the wedge angle determines the deflection of
the prism and not its thickness, the array of little prisms
. could effect the same deflection without the disadvantages of
slow speed and cloudiness caused by large thicknesses. Our past
"" attempts to construct such structures in metal and in glass have
C thus far proven difficult and expensive. However, we have
devised a new idea to solve the problem. We are in the process
of fabricating the periodic structure by using a technique
developed for other purposes at our laboratory: the anisotropic
U etching of silicon for the production of blazed deffraction
gratings. This work was done to much finer tolerances and
smaller dimensions than we require for our application.
Anisotropic etching of single crystal silicon is crystal
p orientation dependent. Silicon has a diamond cubic crystal
structure. The anisotropic etch rate in the (100) direction is
typically greater than the (110), which is greater than the
.(111) direction. The etching process is influenced by the
crystallographic orientation of the substrate, the alignment of
the grating mask pattern to the crystal orientation, and the
etchant chemical.
The fabrication process basically involves a substrate
with the proper crystal orientation to achieve a particular
grating shape or blazed angle. A SiO. layer is thermally grown
. onto the substrate to serve as a chemical mask for the aniso-
tropic etching. Lithographic patterning of the grating is
L.42
.., 21
initially done in a photoresist layer over the oxide. Alignment
I of the grating pattern to the crystal orientation is important
to achieve proper groove shape. This pattern is transferred
into the oxide coatig by acid etching or by reactive plasma
etching. The substrate is then exposed to the anisotropic
I chemical etchant, such as potassium hydroxide (KOH).
Although symmetric grating profiles can be fabricated using
this technique, well controlled blazed profiles are also
achievable. For example, if a SiO 2 mask is aligned along a
(110) dicection of a (100) silicon slice (see Figure 12), V-
shaped grooves are produced with smooth (111) walls. Unlike
isotropic chemical etching, the etch progresses in the (100)
direction and is terminated at a depth of .707 of the oxide
opening width, where the etch front meets the (111) planes,
intersecting the (100) plane at the edge of the mask opening.
* The apex angle of the groove is 109.470. To fabricate blazed
structures, the silicon substrate can be cut and polished to
achieve the desired blazed angle (see Figure 13). NoLt that
apex angles of either 109.47 ° or 70.53 ° are attainable,
depending upon the crystal cut. Recently, we have, on another
program, fully fabricated a short period (two micrometers)
I grating in a single crystal silicon with significantly reduced
flat top defect (Figure 14). This period is several times
smaller than that which would be required for this program.
22
12144-6
100
-502
U
"111 111
110
Figure 12. V-shaped grooves with (111) walls are produced byetching in (100) silicon slice.
12144-7
-' 109.47° 0
100
CSILICONF SURFACE
110
SF1 B e a 70.53b
1100
SILICONSURFACE
L' Figure 13. Blaze angle is determined by
inclination angle between thesilicon surface and the (111)plane.
23
t".". ." " " ' " " "" " " *" " " - " . " ' , . ., ''.-2 '' .' '" "'" --
13645-1
W".
Figure 14. Short period grating in singlecrystal silicon.
24
I . .e q -A I I
* - .-
5 SECTION 3
PERSONNEL
The professional personnel associated with this research
effort at HRL during this period were Bernard H. Soffer,
principal investigator and program manager, Don Boswell, Anna
Lackner, and David Margerum.
25
4 * '* . * . .. ,
SECTION 4
U PUBLICATIONS IESENTATIONS RESULTING FROM AFOSR SUPPORT
A. PUBLICATI,
This sect~tS written publications for AFOSR support
from the initirting date.
1. A. Arrnandaswell, A.A. Sawchuk, B.H. Soffer, andT.C. Straeal-Time Nonlinear Optical Processing withLiquid Cr ' evices," Proceedings 1978 InternationalOptical Cng Conference, London, 153-T58 (Sept-ember
* . 1978).
2. A. Armnand~swell, A.A. Sawchuk, B.H. Soffer, andT.C. Strapproaches to Nonlinear Optical ?rocessing inReal-Timeceedings International Commission for OpticsCongress,I Spain, 253-256 (September 1978).
~:: ~3. A. Armandswell, J. Michaelson, A.A. Sawchuk,
Processing Halftone Screens," 1978 Annual Meeting,Optical 5' of America, San Francisco, October 1978,J. Opt. Si. 68, 1361 (October 1978).
4. A. Armandoswell, A.A. Sawchuk, G.H. Soffer, andT.C. Straew Methods for Real-Time Nonlinear opticalProcessin.78 Annual Meeting, Optical Society ofAmerica, ancisco, California, October 1978, J. Opt.
Soc. Am..61 (October 1978).
5. A. Armand Sawchuk, T.C. Strand, D. Boswell,B.I!. Soff-eal-Time Parallel Optical Analog-to-DigitalConversiotics Lett. 5, 129-131 (March 1980).
6. B.H. Soff-Boswell, A.M. Lackner, A.R. Tanguay, Jr.,T.C. Stra.d A.A. Sawchuk, "Variable Grating Mode LiquidCrystal Dfor Optical Processing," Proceedings Societyof Photo--i Instrumentation Engineers Los AngelesTechnical-sium - Devices and Systems for Optical SignalProcessin 8&1-87, Los Angeles, CA, February 1980.
*7. B.H. Soff Boswell, A.M. Lackner, P. Chavel,.. A. Sawc'C. Strand, and A.R. Tanguay, Jr., "Optical
Computing Variable Grating Mode Liquid CrystalDevices, !edings; Society of Photo-OpticalInstrumeni Engineers Techinical Symposium East - 1980,OpticalC.ng conference 232, 128-136, W'ashington,D.C., Apri0.
27
7": -7 8 7 e 18 v
8. P. Chavel, A.A. Sawchuk, T.C. Strand, A.R. Tanguay, Jr.,3 !and B.H. Soffer, "Optical Logic with Variable Grating ModeLiquid Crystals Devices," Optics Lett. 5, 398-400,September 1980.
9. B.H. Soffer, J.D. Margerum, A.M. Lackner, D. Boswell,A.A. Sawchuk, A.R. Tanguay, T.C. Strand, and P. Chavel,"Variable Grating Mode Liquid Crystal Device for OpticalProcessing and Computing," Molecular Crystals and LiquidCrystals 70, 145-161 (1981).
10. A.A. Sawchuk, T.C. Strand, and B.H. Soffer, "Intensity to. .Spatial Frequency Transformation in Optical Signal Proces-
sing, Proceeding SPIE "Advanced Inst. on Transformations inOptical Signal Processing." (February 1981).
-" 11. A. Armand, T.C. Strand, A.A. Sawchuk, and B.-. Soffer,"Real-Time Parallel Logarithmic Filtering," Opt Lett. 7,451-453, September 1982.
12. A.R. Tanguay, Jr., C.S. Wu, P. Chavel, T.C. Strand,A.A. Sawchuk, and B.H. Soffer, "Physical Characterization
of the Variable Grating Mode Device," 2pt. Eng., 22,687-694 (1983).
13. A.R. Tanguay, Jr., P. Chavel, T.C. Strand, C.S. Wu, andB.H. Soffer "Polarization Properties of the VariableGrating Mode Liquid Crystal Device," submitted to OpticsLetters.
B B. ORAL PRESENTATIONS
This section lists oral presentations at meetings and
conferences describing research supported by this contract from
the initial starting date.
_ 1. A. Armand, D. Boswell, A.A. Sawchuk, B.H. Soffer, andT.C. Strand, "Approaches to Nonlinear Optical Processing
-. - with Liquid Crystal Devices," presented at the 1978International Optic Computing Conference, London (September
-. 1978).
2. A. Armand, D. Boswell, A.A. Sawchuk, B.H. Soffer, andT.C. Strand, "Approaches to Nonlinear Optical Processing in
* - Real-Time," presented at the International Commission forOptics Congress, Madrid, Spain (September 1978).
b| L7
28
K-
- - ** * ** * -. *:D" ~ ~ .**
3. A. Armand, D. Boswell, J. Michaelson, A.A. Sawchuk,B.H. Soffer, and T.C. Strand, "Real-Time Nonliner Process-
U ing with Halftone Screens," presented at 1978 Annual Meet-ing, Optical Society of America, San Francisco, California(October 1978).
4. A. Armand, D. Boswell, A.A. Sawchuk, B.H. Soffer, andT.C. Strand, "New Methods for Real-Time Nonlinear Process-
* ing," presented at 1978 Annual Meeting, Optical Society of-' America, San Francisco, California (October 1978).
5. A.A. Sawchuk, T.C. Strand, A.R. Tanguay, Jr., P. Chavel,D. Boswell, and B.H. Soffer, "Parallel Optical Analog-to-Digital Conversion Using a Liquid Crystal Light Valve,"Workshop on High Speed A/D Conversion, Portland, Oregon(February 1980).
6. B.H. Soffer, D. Boswell, A.M. Lackner, A.R. Tanguay, Jr.,T.C. Strand, and A.A. Sawchuk, "Variable Grating Mode LiquidCrystal Device for Optical Processing," presented at SPIELos Angeles Technical Symposium-Devices and Systems forOptical Processing, Los Angeles, California (February 1980)
7. B.H. Soffer, D. Boswell, A.M. Lackner, P. Chavel,A.A. Sawchuk, T.C. Strand, and A.R. Tanguay, Jr., SPIETechnical Symposium East - 1980 Optical ComputingConference, Washington, D.C. (April 1980).
8. A.A. Sawchuk, T.C. Strand, A.R. Tanguay, Jr., P. Chavel,D Boswell, A.M. Lackner, and B.H. Soffer, "VariableGrating Model Liquid Crystal Light Valves and TheirApplication to Optical Processing," Gordon ResearchConference on Coherent Optics and Holography, SantaBarbara, California (June 1980).
9. B.H. Soffer, D. Boswell, A.M. Lackner, A.R. Tanguay, Jr.,T.C. Strand, and A.A. Sawchuk, "Variable Grating ModeLiquid Crystal Devices for Optical Processing andComputing," Eighth International liquid Crystal Conf.,Kyoto, Japan (June 1980).
10. A.R. Tanguay, Jr., T.C. Strand, P. Chavel, B.H. Soffer, andA.S. Sawchuk, "Theoretical and Experimental PolarizationProperties of the Variable Grating Mode Liquid CrystalStructure," presented at Annual Meeting of Optical Societyof America, Orlando, FL, Oct. 1981.
29
• . -, • ~~~~~~~~...................................... . .............. -. ... ,- ... ." .
11. B. H. Soffer, "Technical Applications of the VariableGrating Mode Effect," presented at 4th Liquid CrystalConference, Tbilisi, Georgia, USSR, Sept. 1981.
12. B.K. Jenkins, A.A. Sawchuk, B.H. Soffer, and T.C. Strand,"Sequential Optical Logic Implementation," presented atAnnual Meeting of Optical Society of America, Tucson, AZOct. 1982.a
13. U. Efron, B.H. Soffer, M.J. Little, and H.J. Caufield, "TheApplication of Silicon Liquid Crystal Light Valves toOptical Data Processing," presented at the SPIE TechnicalSymposium - Advances in Optical Information Processing,(V. 388) LA, January 1983.
14. A.R. Tanguay, Jr., C.S. Wu, P. Chavel, T.C. Strand,A.A. Sawchuk, and B.H. Soffer, "Physical Characterizationof the Variable Grating Mode Liquid Crystal Device,"presented at the SPIE Technical Symposium - AdvancedOptical Information Processing (V. 388) LA, January 1983.
15. U. Efron, B.H. Soffer, and H.J. Caufield, "The Application
" of Silicon Liquid Crystal Light Valves to Optical Data
"* Processing: A Review," presented at Optical InformationProcessing Conference II, NASA, Langley, VA Aug. 1983.
16. B.H. Soffer, "Liquid Crystal Optical Processing," invitedkeynote paper presented at 13th European Solid State DeviceResearch Conference, University of Kent at Canterbury,Sept. 1983.
3
2'
-4
.
-. 30
. .. . " - -. . . .- v -. ", -.- ,-: ,:* -.*o> *, '- , * .- *..:. : ; . *.- . .
APPENDIX
S PHYSICAL CHARACTERIZATION OF THE VARIABLE GRATING MODELIQUID CRYSTAL DEVICE
31
Physical characterization of the variable grating modeliquid crystal device
A. R. Tanguay, Jr. Abstract. The physical principles of operation of the variable grating modeC. S. Wu (VGM) liquid crystal device are described. The VGM device is capable of per-P. Chavelo forming a two-dimensional intensity-to-spatial frequency conversion, which inT. C. Strandt turn allows the implementation of a wide range of nonlinear optical processing
P A. A. Sawchuk and computing functions. The device utilizes certain nematic liquid crystal" University of Southern California mixtures that are observed to form variable frequency diffraction gratings under
Image Processing Institute, and Departments of the influence of an applied bias voltage. Both fundamental and technologicalElectrical Engineering and Materials Science limitations to device performance characteristics are discussed.
* -. Los Angeles. California 90089-0483Keywords: spatial light modulator: variable grating mode; liquid crystal device, optical
B. H. Softer information processing optical devices.
Hughes Research Laboratories Optical Engineering 22(6, 687-694 (November/December 1983).3011 Malibu Canyon RoadMalibu. California 90265
L,;
CONTENTS The variable grating mode liquid crystal device (VGM LCD) 9- 2
I ,transforms input intensities to spatial positions when used in conjunc-
.I Icerdction tion with a Fourier transform lens. The nature of this image trans-e. Fuce description and operational mode formation can be realized in the following manner. The VGM LCD
3- Fundamental origins of the operational properties primarily consists of a photoconductive layer in series with a layer of4 Physical ongin of the vanable grating mode effect nematic liquid crystal mixture. A dc bias voltage is applied across the5 Acknowledgments device to provide a voltage division between the two layers. Within a6 References given image pixel, the input intensity decays the voltage across the
photoconductive layer and correspondingly enhances the voltageI . INTRODUCTION across the liquid crystal layer. The photoconductor thus implementsA- wide variety of one- and two-dimensional operations are necessary an intensity-to-voltage conversion. The nematic liquid crystal mix-for lull-scale implementation of parallel optical processing and ture employed in the device has the unusual property that thecomputing systems. Incoherent-to-coherent conversions are often alignment of the liquid crystal molecules, which is homogeneous inrequired for algorithms involving spectrum analysis and modifi- the quiescent state, exhibits spatially periodic modulation when a3cation. correlation, convolution, and holographic image formation, bias voltage is applied across the layer. This modulation results in aparticularly when the information to be processed is available only in birefringent phase gratingl3 characterized by a spatial frequency thattime-sequential or matrix-addressed raster format. A number of depends linearly on the applied voltage. The effect of the liquidone- and two-dimensional spatial light modulators capable of this crystal layer is thus to implement a voltage-to-spatial frequencyt.pe ofI image transduction are described within this special issue of conversion. If both layers are considered together. the entire device isOptical Engimeenng,, 1-4 as well as in several review articles.'-' thus seen to perform an image-wise intensity-to-spatial frequency
Other. equally important processing and computing functions, conversion, which can be modified to the more general intensity-to-.,uch as logic operations, programmable matrix addressing, binary position transformation by placing a Fourier transform lens behindaddition, linearity compensation, and input-output nonlinearities the VGM LCD. Collimated readout illumination normally incidentie.g.. exponentials, logarithms, power laws, thresholds, level slices, on the device (at a wavelength of photoconductive insensitivity) isand level restoration), have proven particularly difficult to implement. angle encoded within each image pixel by diffraction from eachAll of these functions, on the other hand, can be implemented by induced phase grating and subsequently angle-to-position mappedmeans of some form of intensity-to-position encoding in conjunction by the Fourier transform lens into its focal plane.with either fixed (single function) or programmable (multifunction) This type of process is shown schematically in Fig. 1, in which themasks This general statement follows from the realization that all of input image is assumed to consist of two separate regions of differingthe functions listed above are special cases of data-dependent multi- intensity. The VGM LCD encodes both regions with different spatialplications. in which the input value (e.g.. pixel intensity) selects the frequencies, resulting in separated diffraction orders in the filterappropnate multiplier (e.g., mask location) to obtain the desired (Fourier) plane. Insertion of an appropriate spatial filter or pro-product (e.g.. output intensity), grammable mask (not shown in Fig. 1) into the Fourier plane allows
the separated orders to be selectively modified to implement anyPermanent address Institut dOptique. Universite de Paris sud. BP 43. 91406 Orsay desired data-dependent multiplication or point nonlinearity. In the
Cedex. France reconstructed output image, all regions of equal input intensity are
'Present address I91 Corporaion. 5600 Cotie Rd.. San Jose. CA 9593. modified identically, irrespective of their location in the input imagefield. Thus. all of the data-dependent multiplications are performed
Invited Paper LM.104received July i. 1983 revised manuscnpt received July 28. 198 in parallel. Functional programmability is achieved by replacementaccepted for pubiication Aug.' . 913. received by Managing Editor Aug. 29. 1983 This or reprogramming of the Fourier plane mask, which need only be apaper is a revision of Paper 38.12 which was presented at the SPIE conference on low resolution device with a total number of resolution elements.4dvinces in Optical Information Processing. Jan 20-21. 1983. Los Angeles. CA. Theper presented there appears unrefereed in SPIE Proceedings Vol. 388. equal to the number of gray levels required to be processed.1163 Sociey of Photo-Optical Instrumentation Engineers The overall input-output characteristic of nonlinear function
OPTICAL ENGINEERING / November/December 1983 / Vol. 22 No. 6 / 687
32
A.A"..TANGUAY. WU.. CH'AVEL STRAND. SAWCHUK. SOFFER
*,_- _., ; __.---q OA=
-7 - ~25um
De ,ce
* Fig. 1. Experimental arrangement for demonstration of inten-ity-to-poation encoding by means of an intenaitV-to-spatial frequency conver-sion in a VGM LCD.
SpatialFrequency
Fourier _ VGMTransform LC LV
Plane i
4- 1K ,-(a)
Spatial ________ nu
Iransmttonce I I Intensity
Square Low _ _- - _ OverallDetection Nonlinearity
Output3 Intensity
Fig. 2. VGM nonlinear procesing. The overall input-output characteris-tic can be found by stapping through the auccesaive nonlinear tranafor-mation., including (1) the intnaity-to-apatial frequency conversion. (2)spatial filtering. and (3) intensity detection.
implementation utilizing the VGM LCD is shown schematically in Fig. 3. Polarization photomicrographs of the liquid cryatal domain Pat!Fig. 2 by specifying sequentially the nature of the transformations tarn in an electrically activated call. In a). the polarizer was oriented atfrom input intensity to spatial frequency within the VGM LCD, from 900, and the analyzer at 90". with respect to the grating wave vector. Inspatial frequency to spatial filter amplitude transmittance in the (b). the polarizer was oriented at 900, and the analyzer at 100, withFourier transform plane, and from output amplitude to output respect to the grating wave vector. The unit vector AA denotes theintensity (usually by means of square law detection). The overall direction of quiescent alignment, and ko indicates the direction of thenonlinearity achieved can be easily compensated for the functional grating wave vector.dependences of the separate steps by adjustment of the selectedspatial filter transmittance function.
The VGM LCD has thus far been utilized to perform a wide of these concepts has appeared previously.' -variety of parallel nonlinear point transformations, including level The critical element of the VGM LCD is a thin (4 to 12 ,m) laverslicing,9.''0 2 binary logic functions (AND. OR. NOR. etc.). 0-i 2 and of neratic liquid crystal mixture' 2 that exhibits a periodic modulationfull binary addition (inputs: two addend bit planes and one carry bit of the liquid crystal director, and hence of the index ellipsoid, underplane: outputs: sum bit plane and carry bit plane). II The purpose of application of an electrc field normal to the plane of the layer. BySthis paper is to describe the physical principles of operation of the means of suitable preferential alignment techniques." 12 the quiescent
• variable grating mode liquid crystal device, identify areas of strength state of the liquid crystal is homogeneous (parallel to the plane of theand weakness, and differentiate limitations to current device per- layer). As will be discussed in Sec. 4, this periodic variation of theformance thought to be fundamental in origin from those that are principal axes of the dielectric tensor gives rise to a birefnngent phaseseemingly technological. Section 2 consists of a more detailed descrip- grating characterized by striking and unique optical properties. 1tion of the device, its operating mode, and its operational properties. The grating can be visualized in a polarizing microscope, as shown inThe fundamental origins of these operational properties are examined Fig. 3. by utilizing the birefringence properties of the periodic per-in Sec. 3. in which the natural focus will be the physical mechanism of turbation. Distinct polarizer analyzer combinations give rise tothe variable grating mode effect in neratic liquid crystal mixtures. remarkably different grating images, as can be seen by comparison ofExperimental and theoretical efforts to elucidate the nature of this Figs. 3(a) and 3(b) (see Sec. 4). Furthermore, the grating period is! / ~mechanism are described in Sec. 4, which concludes with several Fis a)nd(b(eeS.4.Futroe heranprodsobserved experimentally to be related inversely to the applied voltageimportant but as yet unanswered questions. across the layer. Above the threshold for domain formation, there-2. DEVICE DESCRIPION AND OPERATIONAL MODE fore, the spatial frequency of the grating is a linear function of the" DCvoltage across the layer, as shown for a variety of nematic liquidA number of important aspects of device construction and device crystal mixtures in Fig. 4.operation are reviewed in this section. A more complete description This voltage-to-spatial frequency transformation can be optically
68 / OPTICAL ENGINEERING / November/December 1983 / Vol 22No 6
33
PHYSICAL CHARACTERIZATION OF THE VARIABLE GRATING MODE UQUID CRYSTAL DEVICE
600 2N140 II I APL, -
& 2N40 CELL A0 2N40 CELL B
' 2N42 0 2N42 CELL C8 2NII CELLE
500 02N25 CELLF *-*- ....A5 X 2N43 CELLG- -
2NII + MERCK NPV G.LASSW MERCK NP
"" "'I~J 400 -43 - LIQUID CRY TA&L [ i . .. 1 )',(,' .. .""W N4
2N4 _ . A
W Fig. 6. Schematic diagram of the VGM liquid crystal device. Current(A devices re read out in transmission at a wavelength of photoconductiveZ 300- insenaitivity.o0C-""l 3. FUNDAMENTAL ORICINS OF THE OPERATIONAL. / PROPERTIES
As was mentioned in Sec. 2, the operational properties of the VGM>" 200- LCD are primarily determined by the variable grating mode effect
200 exhibited by the nematic mixture liquid crystal layer. The current stateof knowledge concerning the physical origin of this unique effect issummarized in Sec. 4. In this section, we outline a number of key
2N42 factors and considerations that affect several important device prop-erties in order to provide both focus and a frame of reference for the
100 I I I lsucceeding section. These device properties include the accessible range10 20 40 60 80 00 120 of spatial frequencies, the number of accessible gray levels, the func-L 2E VL DC tional dependence of diffraction efficiency on applied voltage, maxi-
mum diffraction efficiency, response time, device uniformity, deviceFig. 4. VOM aatlel freqUency aa a function of applied voltage for various input sensitivity, and device operational lifetime.nemati liquid cryta c mixtuage. The accessible range of spatial frequencies extends from the thresh-
old for grating formation at the low end to the onset of dynamicscattering induced by high electric fields at the high end, as shown inFig. 4. For phenyl benzoate mixtures with slightly negative dielectric
addressed by means of a photoconductor placed in series with the anisotropy (<-0.30) such as HRL 2N40.15 this range extends fromliquid crystal layer, as shown in Fig. 5 and described in the previous approximately 200 line pairs/mm to over 600 line pairs/mm. In ordersection. The photoconductive layer employed in devices constructed to avoid overlap of higher diffracted orders from lower spatial fre-thus far is comprised of evaporated or ion-beam sputtered zinc sulfide quencies with lower diffracted orders from higher spatial frequencies,(ZnS), chosen to optimize the impedance match with the liquid crystal the maximum range that can be processed (uniquely assigned to spe-layer (p > 1010° 0 cm). The layer thicknesses employed were of order cific gray levels) spans a factor of two in spatial frequency. For exam-1.5 to 5 /Am. As shown in Fig. 5, the photoconductive and liquid pie, a usable range in HRL 2N40 extends from 300 line pairs, mm tocrystal layers are sandwiched between indium tin oxide (ITO)-coated 600 line pairs mm without order overlap. This accessible range can be1.2 cm thick glass optical flats. The liquid crystal layer thickness is extended by an additional factor of two by utilizing the orthogonal
P" determined by a perimeter Mylar spacer. polarization behavior of alternating diffracted orders, as described inIn operation, a dc bias voltage is applied between the indium tin Sec. 4. Hence, the accessible range of spatial frequencies observed in
- oxide electrodes, of order 40 to 150 V. The input image to be spatial several of the nematic liquid crystal mixtures tested so far is sufficientfrequency encoded is focused on the ZnS photoconductor. produc- for optimized gray scale processing. It should be noted that althoughing image-wise modulation of the local voltage across the liquid the maximum number of resolution elements that can be processed iscrystal layer, thus effecting a parallel intensity-to-spatial frequency linearly proportional to the highest spatial frequency utilized for
-- conversion. The high lateral impedance of the thin film layers allows devices of a given size (see discussion below), use of significantly largerhigh resolution images to be processed with low pixel-to-pixel cross- spatial frequencies begins to place stringent requirements on the
. talk. The device sensitivity is optimized for exposure at blue and Fourier transform lens due to f-number reduction. For example, tonear-ultraviolet wavelengths due to the peak photosensitivity of zinc utilize a spatial frequency range of 600 line pairs mm requires thesulfide in that spectral region. Quasi-nondestructive readout can be output optics to have an f-number less than 1.2 (or less than 2.6 if theaccomplished at wavelengths beyond the photoconductivity edge, lens is displaced off-axis to accept only the positive diffracted orders).such as that of the He-Ne laser (6328 A). Image erasure occurs with This is primarily a pragmatic limitation rather than a fundamental one.removal of the input image, within the dielectric and liquid crystal as VGM effects have been observed at spatial frequencies exceedingrelaxation times of the device (see Sec. 3). To date, all VGM LCDs 1000 line pairs/ mm.'ethat we have constructed have been designed for transmissive read- The number of accessible gray levels that lead to well-separatedout, although reflective readout is possible with incorporation of an diffraction orders in the filter plane is limited by the ratio of theappropriate dielectric mirror. Such a configuration would have the frequency range between VGM harmonics to the object spectrumadvantage of fully separating the reading and writing functions, bandwidth, as shown in Fig. 6. The object spectrum bandwidth is inallowing for increased effective optical gain, turn limited primarily by two effects: spot size due to diffraction from
OPTICAL ENGINEERING / November/December 1983 / Vol 22 No. 6 889
I' - 34rI.
TANGUAY, WU. CHAVEL STRAND, SAWCHUK, SOFFER
i OPTICALsPOURIER
TRANSFORM
vGM /MAGE LOWEST OBJECT
vGM SPECTRUMFREQUENCY
* •Fig. 6. Gray level resolutlon. The number of acce eible gray levels islimited by the redo of the meparation between VGM harmonics to theob spectrunm bendwidth.
finite-sized pixel apertures, and grating imperfections that causelocal deviations from uniform spatial frequency. The first effect is 25 /Im
• "." fundamental and has been treated previously. ' 0 The principal result Fig. 7. Polaization photomicrograph of. VGM liquid crystal layer ahib-of this analysis is the inequality iting a number of grating dliscontinuities Icircled).
bi o 2N . (I)
* in which b is the pixel width, uo is the lowest usable VGM spatial OCfrequency, and N is the desired number of distinguishable gray levels.This inequality requires that the pixel size contain at least 2N periodsof the lowest grating frequency if N gray levels are to be processed.For example, a 256X256 pixel image could be processed with 32distinguishable gray levels on a 50 mm square device with s0 = 300 ma - ii 1 n NC : -
line pairs, mm. An additional restriction arises from scattering effects oadue to grating imperfections, which tend to further increase the sizeof the diffracted orders. The most common type of imperfection z -
! observed in these devices is thejoining or splitting of grating lines, as i I
shown in the photomicrograph in Fig. 7. The origin of these -discli- a &nations" is not at present understood, although the density of occur- ALrence of such imperfections is directly related to the quality of &A
substrate preparation. Z AA typical measurement of the functional dependence of diffrac- -
tion efficiency on the applied voltage across the VGM liquid crystal < -layer is shown in Fig. 8. Since the applied voltage is linearly related tothe induced grating spatial frequenc%', this relationship is illustrative
". of the dependence of the diffraction efficiency on spatial frequency aswell. To first order, the nature of this dependence is not important tothe implementation of optical processing functions since any varia-tion in diffraction efficiency with spatial frequency can be linearizedby insertion of an appropriate multiplicative filter in the focal plane -
of the Fourier transform lens. In any case, the theoretical functional !C . 30 .2 50 6C 2dependence can be derived only from knowledge of the relationship , L'5E
P between the induced orientational angles of the liquid crystal direc-, tor and the applied voltage across the layer. This relationship is Fig. 8. Diffracton efficiency s a function of applied voltage across a- discussed further in the subsequent section. nemnatlc liquid crystal mixture of phenyl benaoates ItR L 2N40). The laver
The maximum diffraction efficiency that can be achieved at a thicknem was approximately 6 rm. an defined by a perimeter Mvlar- given spatial frequency and applied voltage depends fundamentally 4aer.
-' .on the magnitude of the anisotropy in the index of refraction (An -ne - no, with ne the extraordinary refractive index for polarizationparallel to the molecular axis and no the ordinary refractive index for order is larger than the first diffracted order due to the peculiarpolarization perpendicular to the molecular axis), the magnitude of nature of the birefringent phase grating formed by the VG M distor-the periodic angular reorientation of the liquid crystal director, and tion (details are given for this phenomenon in Sec. 4).the thickness of the VGM liquid crystal layer. Full periodic reorien- The response time of the cell is a critical parameter that directlytation of the index ellipsoid from homogeneous (parallel to the affects the achievable overall processing throughput rate. At presentsubstrate) alignment to homeotropic (perpendicular) alignment for it is the major factor inhibiting widespread incorporation of VGMH R L 2N40 (An = 0.15) in a 6 Mm cell read out at 6328 A gives rise to devices in optical processing systems. The rise time for grating for-an optical phase modulation of approximately 9 rad. Hence. the mation from belowto above threshold varies from mixture to mix-
- maximum diffraction efficiency is fundamentally limited to that ture. but is typically of order one second. The response time forexpected for a pure sinusoidal phase grating." In practice, full grating change in response to a step increase in applied voltagereorientation is typically not achieved before the onset of dynamic (corresponding to a step increase in grating spatial frequency) isscattering, although reorientation angles of 450 are thought at pres- typically of order a fraction of a second. A The response time of theent to be commonly reached (see Sec. 4). As can be seen from Fig. 8. photoconductive voltage division across the liquid crystal layer is nottypical second-order diffraction efficiencies are of order 20%. This a significant factor by orders of magnitude relative to the reorien-
690 'OPTICAL ENGINEERING / November/December 1983 /Vol. 22 No. 6
35
PHYSICAL CHARACTERIZATION OF THE VARIABLE GRATING MODE LIQUID CRYSTAL DEVICE
tation response time, so that eventual improvements in VGM LCDresponse time will accrue only by advances in the state of understand- C %1 ------.
ing of the physical origin of the VGM effect and the dynamical natureof the grating reorientation process, followed by appropriate modifi- -YLARcation of the device operational mode to enhance the rate of molecu- SPACER
,' lar reorientation and /or a search for a nematic liquid crystal mixture," with physical characteristics optimized for dynamic VGM effects.
The input sensitivity of the VGM LCD, defined as the input VGM IQUiD% (writing) intensity per unit area per unit change in grating spatial CRYSTAL LAYER -
- " frequency, is determined by a number of factors. These include the*slope of induced grating spatial frequency as a function of applied
voltage for the particular nematic liquid crystal mixture employed,the wavelength dependence of the photoconductive layer photosensi-tivity, and the cell switching ratio (fractional increase in voltage -
*. across the liquid crystal layer from illumination at the threshold for " ..
.- grating formation to saturation). The first factor (liquid crystal- • response slope) varies significantly from mixture to mixture (see Fig.
4). Of the VGM nematic liquid crystal mixtures investigated to date,HRL 2N40 has proved to be nearly optimum in this regard. It is not ITO TRANSPARENT -*- . -.
yet clear what fundamentally influences and eventually limits this COUNTERELECTROOESparameter. The photosensitivity of the photoconductive layer is
- determined primarily by the choice of photoconductive material Fig. 9. Variable grating mode liquid crVKa test geometry showing theCartesian coordinate system refened to in the text a well as the molecu-
(limited to those that can be appropriately impedance-matched to ler orientation angles a and n. This configuration was utilized in thethe liquid crystal layer), method and quality of thin film deposition. polarized Ilght diffraction ,effcincy and photomicroecopy experiments.
"- layer thickness, spectral width and central wavelength of the expo-sure (writing) illumination, and the operational bias voltage em-ployed. The cell switching ratio is a function of the series impedanceof the liquid crystal layer, the impedance of the unilluminated photo- P IN P DIFFRACTEDconductive layer, and the impedance of the photoconductive layer
* . under saturation illumination. In addition, the cell switching ratio"-" will be altered by incorporation of surfactant layers to improve liquid
* crystal quiescent alignment, and of a dielectric mirror in the reflective ' 1 2 3..readout device structure. At this stage of the device development, theinput sensitivity of the VGM LCD has not been optimized. A typical11 £ Avalue of IS (MW: cm 2) - (mm, line pair) was obtained with a VGM I l-l,".,LCD consisting of a 6Am layer of HRL 2N40 in series with a 5 m 1 2 3 4thick evaporated ZnS layer that had been polished and rubbed with
..." surfactant polyvinyl alcohol, operated at 160 V dc, and illuminated VGM DOMAINin the passband 410 to 550 nm.9 ORIENTATION I
The uniformity of VGM LCD response depends inherently on 1 2 3 4technological issues, including uniformity of layer thicknesses, homo-
* geneous mixing of the liquid crystal material employed, and the* as-deposited spatial dependence of photoconductive sensitivity.
Whereas it is relatively straightforward to construct electrically acti- 1 2 3 4vated VGM cells (see Sec. 4) that exhibit a high degree of spatial I
. uniformity, deposition of a photoconductive layer with equivalent Fig. 10. The polarization behavior of VGM diffracted orders, with illumi-spatial homogeneity has proven more difficult. Nonuniformi ty of the nation normal to the plane of the liquid crystal layer. The Ieft-hand columndevice response characteristic can be a contributing factor in the indicates the input polarization associatedwith each rowof output polari-establishment of the maximum number of accessible gray levels zations. The inset shows the corresponding orientation of the VGMdiscussed previously. In experimental devices constructed thus far. grating.
device uniformity has not proven to be the limiting factor. In anycase. it is expected that response nonuniformities can be minimizedsignificantly by improvements in the photoconductive layer deposi- reorientation process. The elucidation of the fundamental nature oftion process. the grating (molecular orientation angles as a function of applied
The lifetimes of experimentally constructed VGM LCDs have voltage across the liquid crystal layer) and of the physical mechanismranged from less than a week to over a year. The causes of VGM that gives rise to the observed periodic instability and its reonenta-device failure have not yet been extensively studied, although several tion dynamics has been a subject of considerable experimental and
contributing factors can be identified. These factors include the theoretical interest. ' V1. - 30 In this section the present state ofpurity and composition of the liquid crystal mixture employed, the understanding of the VGM effect is described, and a number ofnature of the liquid crystal/ photoconductive layer interface, the important unresolved questions are presented.integrity of the device sealing process, and the device operational The grating produced by the periodic spatial reorientation of thehistory. Since the VGM effect requires a dc applied voltage, unidirec- nematic liquid crystal molecules is quite unusual, giving rise totional ion poisoning may contribute to gradual device degradation. siking polarization-dependent properties. These diffraction effects
can be investigated in an electrically activated VGM cell with noPREintervening photoconductive layer, as shown in Fig. 9. The orienta-
4. PYSIAL OIGI OFTHE ARIBLEtion of the grating is such that the grating wave vector is perpendicu-GRATING MODE EFFECT lar to the direction of unperturbed alignment, which is homogeneousAs can be clearly understood from the discussion presented in Sec. 3, a and induced by unidirectional rubbing or ion-beam milling. That is,vast majority of the important device operational properties depend the periodic modulation direction is perpendicular to the initial (zerocritically on the detailed nature of the grating formation and dynamic applied bias) liquid crystal director (long molecular axis), as shown
OPTICAL ENGINEERING / November/December 1983 /Vol. 22 No. 6 / 69136
TANGUAY, WU, CHAVEL STRAND. SAWCHUK. SOFFER
in the polarization micrographs (Fig. 3). 2rxFor all linear input polarization angles, the even and odd diffrac- a = Cos -
ton orders are found to be essentially linearly polarized. In addition. (4)
.. the even diffraction orders are nearly linearly polarized parallel tothe "domains" comprising the VGM grating, as shown in Fig. 10. For 2 7rx
input polarization perpendicular to the domains. the even orders are 7 = o sin7" . found to be almost fully extinguished. On the other hand. the odd
diffraction orders are nearly linearly polarized with a major axis that This procedure allows the extraction of the maximum onentationalrotates counterclockwiseat thesame rateas theinput polarization is excursion angles ao(V) and no(V) as functions of the applied biasrotated clockwise. This effect is the same as that produced by a voltage above the threshold for grating formation, as shown in Fig.
1 half-wave plate oriented at 450 with respect to the grating wave 12(ao) and Fig. 13(770). In each case. subject to the assumed forms ofvector. For input polarization at 450 to the wave vector, all orders are a and v7 implicit in Eq. (4). it is observed that the maximum excursionobserved in the far-field diffraction pattern. An analyzer placed on angles both in and out of the plane of the grating seem to increase asthe output side of the VGM device can be rotated to extinguish the the loganthm of the applied voltage.even orders (when oriented parallel to the grating wave vector) or the The spatial distribution of the ends of the liquid crystal moleculesodd orders (when oriented at -450 to the grating wave vector).Thes unsualpolriztionproertis hve ecenly een tilzed described by Eq. (4) is approximately cycloidal. Such a dependence oft dtThese unusual polarization properties have recently been utilized the angles a and ,1 on the spatial coordinate x has been predicted byto determine the spatial disrbution of the molecular onrentation direct minimization of the free energy in a similar nematic liqud crystalof the diffraction phenomena is directly related to the formation of a system.2' This particular solution is obtained by incorporation of theotefrion phenomen is dic the foia a io of a converse flexoelectric effect in the expression for the free energy. 1 31- ' 3birefringent phase gr'ating" in which the principal axis of the index The flexoelectric effect describes a strain-induced polarization thatellipsoid vanes periodically both in the plane of the grating (charac- arises due to molecular shape effects in conjunction with a nonzeroteized by an orientation angle a) and normal to the plane of the dipole moment, as shown schematically in Fig. 14. The conversegrating (characterized by an onentation angle Y7). The angular coor- flexoelectric effect thus pertains to a polarization-induced strain withindinates are as shown in Fig. 9.dinaes re a shwn , Fi. 9the liquid crystal laver, which can result in a periodic molecular re-
The polarization properties of light diffracted by the liquid crystal thenlii craaerich a resin riodicnmoecular rebirerinentphae gatigca besummrizd b mens fa ranfer orientation characterized by a linear dispersion relation between the
birefringent phase grating can be summarized by means of a transfer grating wave vector and the applied field, as is observed expenmen-matrix that connects the output polarization at each point (x.y) on tally. Including the dielectric, distortion, and flexoelectnc contnbu-
* the rear surface of the liquid crystal layer with the input polarizationat the front surface of the liquid crystal layer. On the basis of the tions to the free energy, yields an expression of the form
expenmental observations, this matrix must be of the form
0 [ AFVp) B(x;2p) FVGM = (Fdielctnc + Fdistonrion + Fflexoelectnc 1dV
(2)(x2p) Do- D(xp) F0 - J(t, - to) (.E )-dV
in which the notation A(x-p) indicates that the complex amplitude A 1fFvaries in the x direction with periodic repetition distance p. For + K 1(7 -) 2 'K,(ii X-uniaxial liquid crystal molecules at an arbitrary orientation (a.??). 2fassumed uniform throughout the layer thickness at a given coordi-nate in the x direction, the Jones matrix can be determined by ( 7 Xappropriate rotations of the index ellipsoid, which yields dV
-sin.( I 00) sinacosaf(I 1- e,imn ) .
Linacosal -ejol I -Cosa( -eJ16J e-3 [(7 X )X ] dV (5)
where- in which K1, K, and K3 are the elastic constants for splay, twist, and
2 rt C0112~ bend deformations, respectively: n is the liquid crystal director: E is- ----.-- -- 0--no the applied electric field; t ande are principal components of the
dielectric tensor of the liquid crystal: and e, and el are flexoelectnccoefficients.Y1 Minimization of FVGM with respect to the onentation
in which t is the liquid crystal layer thickness, no is the ordinary index angles a and 17 of the director, subject to fully pinned boundary
of refraction. ne is the extraordinary index of refraction. and X is the conditions at both substrate surfaces, and with the simplifyingwavelength of readout illumination employed. The angles a and 77 assumption that Ki = K2 = K. yields3'are assumed to be periodic functions of x and independent of y and z.Measurement of the intensities in each diffraction order for a mm- a = aocos( kx)cos( Yrz t) .(6)imum set of polarizer analyzer orientations uniquely determines the n = xmagnitudes of the Fourier components of the polarization transfer 7 = ?oSin( kx)cos( inz t)
* ', matrx. These experimentally derived values can then be compared in which k is the grating wave vector, and t is the liquid crystal layerwith the theoretically calculated coefficients of Eq. (3) (by harmonic thickness. This solution generates a dispersion relation between Eexpansion) for different posuble assumptions concerning the spatial an the fosdistribution of the orientation angles a and Y7. An example of such a and k of the form
.ah e k:. comparison between theory and experiment is shown in Fig. II (K 4,(r.t)2 2
under the assumption that the spatial dependences of a and n~ are E2 . 4)
* given by e)' k - [k" 1, (1r:t)
9-;. g692 / OPTICAL ENGINEERING / Novumber/Oeembsr 1983 / Vol. 22 No,.6
- 3"7
• """" """ " "" - " """"" " """ - " " .. ', "": " """ '",.. '',.::,.:. .:-.-::. "':,''.:'-....-.- ... , *.-',-" ,'..v" ,'-"" """"""" ""'""" """""""-""""""""
PHYSICAL CHARACTERIZATION OF THE VARIABLE GRATING MODE LIQUID CRYSTAL DEVICE
40 +17DEG)
0 30
D.FqAC'ED ORDER NTENSI!T:SvGAA C:EL. KC74. '23A
APP 32v 25 VGM CELL K1074- 123A
5C C' 00 0L~*.-E-QEr CAL DATA 0. lc10
APPLIED VOLTAGE fV) 0
Fig. 11. Meaeured diffracted order intensities ae function of theoreticalintensitiee calculated from the uniaxial VGM model described in the text. Fig. 13. The out-of-plane molecular orientation angle as ea function of
the applied dc bias voltage scroaa the call IV) (see Fig. 5).
25 SUBSTRATEN 2~ DEG;
P 0
20 4~U I SUBSTRATE
SUBSTRATE
++ +
P 0VG'M CELL K1074-123Afcall '6 Open UBS R
Fig. 14. Schematic diagram illustrating a possible mechanism for theoccurrence of the flexoelectric ee due to a shape anisotropy in liquid
I I crystal molecules with a permanent dipole moment. In tel. the molecular0--tq orientations awe randomly distributed, resulting in zero net polarization. In
100 (b) the applied distortion inducesn a shapes-dependent molecular realign-APPLIED VOLTAGE IV) ment. reeulting in a not polarization. lAfter Ref. 31.)
Fig. 12. The in- plane molecular orientation angle I ao) as a function of theapplied dc biaa voltage across the call IV) Ieass Fig. 5). not yet been established beyond doubt. although the accumulated
evidence points strongly toward the converse flexoelectnc effect.This assignment is also intuitively appealing since the VGM effect is
in which e* - el - e3, iIA M (*aK. 41re*2 ), and ea m te - to. The observed only in liquid cry~tal mixtures with slightly negative dielec-dispersion relation is linear when k > 7r. t, as shown in Fig. 15 tric anisotropy. 12 for which both the dielectric and distortion contni-(compare with the experimental relationship shown in Fig. 4). butions to the free energy increase for deviations from uniform
The origin of the periodic instability in VGM liquid crystals has alignment. The addition of the flexoelectric term counteracts these
L OPTICAL ENGINEERING,, November/IDecember 1983 / Vol.22 No. 6 / 693
38
TANGUAY. WU. CHAVEL. STRAND. SAWCI4UK. SOFFER
only one possible means of achieving parallel intensity-to-posit ionencoding. The processing potential of the intensitv.to-p'osition algo-I CMAi , A -. . -nES-X;: 4rithm should provide more than adequate inducement to intensitYthe search for alternative implementations.
S . ACKNOWLEDGMENTS* .. -The authors gratefully acknowledge the contributions of D Boswell.
A. M. Lackner. andiJ. D. Margerumn Hughes Research Laboratories).- and K. Sherman and G. Edwards U niversitv of Southern California i
RA7-NG FMA Q This research was supported by the U.S. Air Force Office of ScientificP Research. Electronics and Solid State Sciences Division. under grant
AFOSR-8 1-0082 at the U rnversity of Southern California and contract
30C -;:c 5ccF49620-8 I-C-0086 at Hughes Research Laboratories..GM O~N bPr ., ~ "e O'5~m~6. REFERENCES
Fig. IS. Reutn hoeeidseptnrlgo e iudeV am I .R. V. Johnson. D. L. Hecht. R. A. Sprague. L. N Flores. D L Stein-fero I Cftao W4 to HFIL 2N40 phenyl bertoase mbitur (ane Eq. (7)), metz. and W D. Turner. Opt. Eng. 22216), 665 1 98 3subi.a to the ammmPtion that K, = K, = IC. 2. D L. Pape and L. J. Hornbeck. Opt. Eng. .216). 675 1983)
3. U. Efron. P.O. Braatz. M. J. Little. R N Schwartz. and J Grinberg.Opt. Eng. 22(6). 682 (1983).
4, C. Warde and J. Thackara. Opt. Eng. 2216). 695 1983)effects, producing a free-energy minimum at nonzero distortion 5. 0. Casasent. Proc. IEEE 63. 14311977),angles. Furthermore, this model predicts a static periodic perturba- 6. S. Lipson. 'Recyclable Incoherent-to-Coherent Image Conv.erters.-ition; no hydrodynamic or electrohydrodynamic effects are observed Advances in Holography. N Farhat. Ed.. Marcel Dekker. New N orK
in or tet cels.11979).in ou tes cels. 7G. Knight. -Interface Devices and Memory Materials'* in Optical DataThe dynamics of grating formation, reorientat ion. and relaxation Processing. S. H. Lee. Ed.. Springer-Verlag. Heidelberg t 97611*are subjects of ongoing experimental and theoretical inetgain 8. A. R. Tanguay. Jr., Proc. ARO Workshop on Future Directions tor
%Understanding of the basic physical principles underlying these Optical Information Processing. Texas Tech University. Lubbock.effects is vital to the success of efforts to improve the VGM LCD 9. B. H. Soiffer. D. Boswell. A. M. Lackner. A. Rt. Tanguay. Jr.. T Cresponse time. Strand. and A. A. Sawchuk. in Devices and Svutems 'for Optical Signal
A number of spatial discontinuities which are similar in appear- Processin. T. C. Strand and A. It. Tanguay. Jr.. Eds.. Proc. SPIE 218.ance to crystallographic dislocations can be observed within the 908)
liqid rysal aye (se Fg. ). he islcaton~ desit inreaes 10, B. H. Soffer, D. Boswell. A, M. Lackner. P. Chavel. A. A. Sawchuk. Tliqud cystl lyer(seeFig 7) Th 'dsloatio- dnsiy icresesC. Strand, and A. R. Tanguay. Jr.. in 1980 International Optical Comn-with timeas each VGM device deteriorates. and strongly depends on puting Conference (Book 1I), Proc. SPIE 232. 12801980).Uthe manner in which the applied bias is brought from below to above U. P Chavel. A. A. Sawchuk. T. C. Strand. A. R. Tanguay. Jr . and B Hthe hrehol ofgatig frmaton.In omecses asthe ppled iasSoffer. Opt. Lett. 5. 398(19801.thehreholofgatigfomaton.In omease~asheaplid bas 2. B. H. Soiffer. 1. D. Margerum. A. M4. Lackner. D Boswell. Ai R.across an electrically activated cell is increased or decreased, the Tanguay, Jr.. T. C. Strand. A. A. Sawchuk. and P Chavei. Mo). Cryst.
dislocations propagate past each other while the grating period Liq. Cryst. 70. 14501981).decreases or increases, respectively Dislocations with opposite onen- 13. A. R. Tanguay. Jr . submitted for publication in Opt. Lett.
4. A. R. Tanguay. Jr.. P. Chavel, T C Strand. C. S Wu. and B. H Soifer.tations propagate in opposite directions until a new stable equilib- submitted for publication in Opt. Lett.nrn situation is achieved. It is not yet clear whether these disloca- is I B.Margerum.J. E. Jensen. and A.M. Lackner.Mol. Crvst. Liq. Crs.* lions represent true disclinations. for which the liquid crystal director 68. 137(181),
is discontinuous near the defect. or whether they are in fact alterna- 16. W Greubel and U Wolff. AppI. Phys. Lgti 19('!. 2 13419711tive~~~~~~~~~~~~~ lclcniuusouintotemnmztopr I' J, W Goodman. Introduction to F ourier Oputcs. McGraw-Hill. Newfie ocl otiuos olton t tefree-energy miiiain e- York (1968).haps induced by an as yet undiscovered perturbation. 18. B. H. Soffer. -Real-Time Implementation of Nonlinear Optical Process-
large number 0f research directions have been described in the ing Functions." Annual Technical Report an AFOSR F496.0-81-C.foregoing paragraphs. In addition, several others are worthy of note. 0086.B. M982). hlReLtt2(8.1199First. numerous experimental measurements of the off-diagonal (90. . eer.nk ,PAws. etrv LC h P2(? Khio& an1BL969 ~sielements in the Jones matrix describing polarized light propagation Bulg. J Phys. 2. 16501974),
-through the VGM cell have revealed a significant asymmetry not 21. Yu. P Bobylev and S. A. Pikin. Sov Phi's JETP 450(1. 195i 19""predicted by theuntaxial model. The orspoof this -BiC-aymmetry 22. Yu. P Bobvlev. V G. Chigrinov. and S. A. Pikin. J de Phys 40. Suppi.
-. 4. CJ-331 61979).effect has not yet been elucidated. Second. the thicknessadependence of 23. S, A. Pikin. Mol. Crvst. Liq. Cryst. 63. 181119811the molecular orientation angles a and np has not been fully established. 24 Pla 8 n . Flannery. Society for Information Display It976The solution proposedi in Eq. (6) assumes full surface pinning at the Intern. Ss'mp. Digest. 143(1976i.substrate boundaries, and represents the lowest order z-dependent 25. J M. Pol lackandi B. Flannery.in Liquid C" stals and Ordered Fluid..mode TheJ. F Johnson and R. E. Porter. Eds.. Vol 2. pp 551,-57 1. Plenum Press.md.Tepolarization-dependent diffacted order meaurementsNeYoki97.
described herein do not provide clear differentiation between a uni- 2 6. J, M. Pollack and 3. B. Flannery. in Liquid Cr sals and Orderred Fluid..form z-dependence and the lowest order mode. Third. the nature of J F Johnson and ft. E. Porter. Eds.. Vol.- 3. pp 4 21 -1442. Plenum Press.the transient solutions to the free-entergy minimization incorporating New York 11978).molecular dynamic and viscosity effects has not been treated. Solu- 27L K iti' SyP ysalokl 1511). 5 l1k190tion of this problem iskeyto response time optimization of the VGM 29 A. Derzhanski. A.G. Petrov. and M. D Mitov.J dePhxs 3.'i8liquid crystal device. Fourth. it is not yet clear whether a nematic 30. M 1. Barnik. L. M Blinov. A'. N Trufanov. and B A U .manski. J de
* liquid crystal mixture can be found that exhibits a useful VO M effect Phys. 39. 41711978)under ac bias. Such a mixture may provide significant immunity 31 P G. de Gennes. The Phissics of Liquid Crvsials. pp 4'-101. Owford
Iofrom long-term ion-poisoning effects. Finally, it should be emnpha'- .32 Uiversity Press. London 11974).32J Prosi and J P Marcerou. J die Ph'.s 38. 315i19'")sized that the variable grating mode liquid crystal effect provides 33. J Prost and P S. Pershan. J Appl. Phvs. 47. .1298(19161
694 'OPTICAL ENGINEERING / Novtember 1Oecemyber 1983 / Vol, 22 No. 6
4141
k -A
r -A, -
Ar
4 '4
'I4* ~~8
IVtPUb
.U~~-~~ .';
"Ilk.
Recommended